ReplicationWorkerHelper OOM (Java heap space) in orchestrator container - JAVA_OPTS not configurable

### Helm Chart Version

2.0.2-alpha-c905e75

### What step the error happened?

None

### Relevant information

## Description
The orchestrator container in replication pods consistently fails with
`OutOfMemoryError: Java heap space` in `ReplicationWorkerHelper.internalProcessMessageFromDestination`.

## Observed behavior
The orchestrator container has `JAVA_OPTS` explicitly set to empty in the pod spec,
and there is no helm values mechanism to configure it.

Steps to reproduce:
1. Run a sync with a Java-based destination connector under load
2. Observe OOM in orchestrator container logs:
`Terminating due to java.lang.OutOfMemoryError: Java heap space`
at `i.a.c.o.w.ReplicationWorkerHelper(internalProcessMessageFromDestination):312`

## Investigation findings
- The orchestrator container has `JAVA_OPTS` with empty value (hardcoded by workload-launcher)
- `JAVA_TOOL_OPTIONS` is NOT injected into the orchestrator container
- `global.env_vars.JAVA_TOOL_OPTIONS` and `global.env_vars.JVM_ARGS` do NOT propagate to the orchestrator
- `workload-launcher.extraEnv.JAVA_OPTS` is set on the launcher pod but NOT propagated to the orchestrator
- Pod memory limit: 16Gi — pod consuming ~7G at time of failure
- `CONTAINER_ORCHESTRATOR_ENABLED=false` on the workload-launcher

## Expected behavior
A helm values mechanism to configure `JAVA_OPTS` for the orchestrator container,
e.g. `workload-launcher.orchestratorJavaOpts` or via `CONTAINER_ORCHESTRATOR_JAVA_OPTS`.

## Environment
- Kubernetes: AWS EKS
- Deployment: Helm Chart V1 (OSS community)

### Relevant log output

```shell

```

---
**Internal Tracking:** https://github.com/airbytehq/oncall/issues/13381